A robust, multi-solution framework for well placement and control optimization

Field development and control optimization aim to maximize the economic profit of oil gas production while considering several sources uncertainty. This results in a high-dimensional problem with computationally demanding uncertain objective function based on simulated reservoir model. The limitations many current robust methods are: 1) it is single-level (e.g. well locations/placement only; or production/injection variables only) that ignores interference between from different levels; 2) they provide single optimal solution, whereas operational problems often add unexpected constraints likely reduce optimal, inflexible solution sub-optimal scenario. paper presents robust, multi-solution framework sequential iterative at multiple levels using Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm. A systematic realization selection process, tailored subsequent stage, used select small representative ensemble model realizations be for calculating expected value. estimated gradients are calculated 1:1 ratio mapping perturbations each iteration onto selected computational cost. An close-to-optimum solutions then chosen level placement level) transferred next where settings optimized), this loop continues until no significant improvement observed Fit-for-purpose clustering techniques developed systematically an solutions, maximum differences but values, level. proposed has been tested benchmark case study (Brugge field). Multiple obtained locations values. We show suboptimal early can approach even outdo one level(s). Results demonstrate advantage more efficient exploration search space providing much-needed flexibility field operators.